Fix TSV reader losing empty string values in provenance/metadata/rescoring dicts

CHANGELOG.md

@@ -5,6 +5,13 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [1.5.3] - 2026-05-13
+
+### Fixed
+
+- `io.fragpipe`: Fix parsing of modifications (support both splitting on `,` and `, `).
+- `io.tsv`: Fix empty string values in `provenance_data` and `metadata` being lost (converted to `None`) when reading TSV files; empty rescoring feature values are now parsed as `NaN` (fixes #145).
+
 ## [1.5.2] - 2026-02-12
 
 ### Fixed

psm_utils/__init__.py

@@ -1,6 +1,6 @@
 """Common utilities for parsing and handling PSMs, and search engine results."""
 
-__version__ = "1.5.2"
+__version__ = "1.5.4"
 __all__ = ["Peptidoform", "PSM", "PSMList"]
 
 from warnings import filterwarnings

psm_utils/io/tsv.py

@@ -51,6 +51,7 @@
 import ast
 import csv
 import logging
+import math
 from collections.abc import Iterator
 from pathlib import Path
 from typing import Any, TextIO
@@ -110,14 +111,14 @@ def _parse_entry(entry: dict[str, str | None]) -> dict[str, Any]:
         parsed_entry: dict[str, Any] = {}
         provenance_data: dict[str, str | None] = {}
         metadata: dict[str, str | None] = {}
-        rescoring_features: dict[str, str | None] = {}
+        rescoring_features: dict[str, Any] = {}
         for k, v in entry.items():
             if k.startswith("provenance:"):
-                provenance_data[k[11:]] = v
+                provenance_data[k[11:]] = v if v is not None else ""
             elif k.startswith("meta:"):
-                metadata[k[5:]] = v
+                metadata[k[5:]] = v if v is not None else ""
             elif k.startswith("rescoring:"):
-                rescoring_features[k[10:]] = v
+                rescoring_features[k[10:]] = v if v is not None else math.nan
             else:
                 parsed_entry[k] = v
 

tests/test_io/test_tsv.py

@@ -1,5 +1,7 @@
 """Tests for psm_utils.io.tsv."""
 
+import math
+
 import pytest
 
 from psm_utils.io.exceptions import PSMUtilsIOException
@@ -27,13 +29,44 @@
             "rescoring_features": {},
         },
     ),
+    (
+        # Empty string provenance value (e.g. missing optional provenance field)
+        {"peptidoform": "ACDE", "spectrum_id": "1", "provenance:missing": ""},
+        {
+            "peptidoform": "ACDE",
+            "spectrum_id": "1",
+            "provenance_data": {"missing": ""},
+            "metadata": {},
+            "rescoring_features": {},
+        },
+    ),
+    (
+        # Empty string rescoring feature value should become NaN
+        {"peptidoform": "ACDE", "spectrum_id": "1", "rescoring:score": ""},
+        {
+            "peptidoform": "ACDE",
+            "spectrum_id": "1",
+            "provenance_data": {},
+            "metadata": {},
+            "rescoring_features": {"score": float("nan")},
+        },
+    ),
 ]
 
 
 class TestTSVReader:
     def test__parse_entry(self):
         for test_in, expected_out in test_cases:
-            assert TSVReader._parse_entry(test_in) == expected_out
+            result = TSVReader._parse_entry(test_in)
+            for key, expected_val in expected_out.items():
+                if isinstance(expected_val, dict):
+                    for k, v in expected_val.items():
+                        if isinstance(v, float) and math.isnan(v):
+                            assert math.isnan(result[key][k])
+                        else:
+                            assert result[key][k] == v
+                else:
+                    assert result[key] == expected_val
 
     def test_iter(self):
         reader = TSVReader("tests/test_data/test.tsv")